Salifort Motors Employee Retention Dashboard

Executive Summary

Executive Summary: Employee Retention at Salifort Motors

Project Overview

  • Developed a machine learning model to predict employee turnover at Salifort Motors
  • Achieved 98% accuracy and 0.97 F1 score using Random Forest classifier
  • Identified key factors contributing to employee turnover

Key Findings

  • Employee satisfaction level is the strongest predictor of turnover
  • Employees with high number of projects and long working hours are at risk
  • Department and salary level have significant impact on retention
  • Employees who stayed 3-4 years with no promotion show higher turnover

Recommendations

  • Improve Satisfaction: Implement regular satisfaction surveys and feedback mechanisms
  • Workload Balance: Review project allocation and working hours to prevent burnout
  • Career Development: Create clearer promotion paths, especially for employees after 3 years
  • Compensation Review: Evaluate salary structures in departments with high turnover
  • Recognition Programs: Develop meaningful recognition for employee contributions

Implementation Timeline

  • Immediate (0-3 months): Satisfaction surveys, workload review, recognition programs
  • Short-term (3-6 months): Career development programs, compensation adjustments
  • Long-term (6-12 months): Comprehensive retention strategy, leadership training
Model Comparison

Performance Metrics of All Models

Model Accuracy F1 Score Precision Recall
Random Forest 0.970 0.980 0.960 0.950
Logistic Regression 0.830 0.760 0.800 0.730
Decision Tree 0.950 0.940 0.930 0.920
XGBoost 0.960 0.950 0.940 0.930
Model Comparison
Comparison of performance metrics across all tested machine learning models.

Why Random Forest Was Selected

  • Highest F1 score (0.98) among all models
  • Excellent balance of precision and recall
  • Superior performance in cross-validation tests
  • Better generalization to unseen data
  • More resistant to overfitting compared to Decision Tree
Confusion Matrix
Confusion matrix showing true vs. predicted values for the Random Forest model.
ROC Curve
ROC curve illustrating the diagnostic ability of the Random Forest classifier.
Feature Importance
Feature Importance
Feature importance plot showing the relative importance of each feature in predicting employee turnover.
Top 5 Important Features
satisfaction_level, number_project, time_spend_company, average_montly_hours, last_evaluation
Key Findings
Department Distribution
Distribution of employees across departments and salary levels.
Target Distribution
Proportion of employees who stayed vs. those who left the company.
Satisfaction Distribution
Distribution of employee satisfaction levels, a key predictor of turnover.
Working Hours
Relationship between average monthly working hours and number of projects.
Satisfaction vs Evaluation
Relationship between satisfaction level and last evaluation score.
Time at Company
Distribution of time spent at the company, showing turnover patterns by tenure.
Correlation Matrix
Correlation matrix showing relationships between all numeric variables.
Recommendations
Immediate Actions (0-3 months)
• Implement satisfaction surveys and feedback mechanisms
• Review and adjust workload distribution
• Enhance recognition and reward programs
Short-term Initiatives (3-6 months)
• Develop career development programs
• Improve work-life balance policies
• Enhance training and development opportunities
Long-term Strategies (6-12 months)
• Implement comprehensive retention programs
• Develop succession planning
• Create mentorship programs